<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>GoldFinch: High Performance RWKV/Transformer Hybrid with Linear Pre-Fill and Extreme KV-Cache Compression · HackerPapers discussions</title><link>https://hackerpaper.kaipfister.de/p/2407.12077/</link><description>Collected HN comments and submissions linking this paper.</description><language>en</language><lastBuildDate>Mon, 02 Nov 2026 11:16:22 GMT</lastBuildDate><atom:link href="https://hackerpaper.kaipfister.de/p/2407.12077/feed.xml" rel="self" type="application/rss+xml" /><item><title>GoldFinch: High Performance RWKV/Transformer Hybrid with Linear Pre-Fill and Extreme KV-Cache Compression — mentioned by pico_creator</title><link>https://news.ycombinator.com/item?id=42572451</link><guid isPermaLink="false">urn:hackerpapers:mention:2407.12077:42572451</guid><pubDate>Thu, 02 Jan 2025 07:19:28 GMT</pubDate><description>&lt;p&gt;Not an MoE, but we have already done hybrid models. And found it to be highly performant (as per the training budget) https://arxiv.org/abs/2407.12077&lt;/p&gt;&lt;p&gt;&lt;a href="https://hackerpaper.kaipfister.de/p/2407.12077/"&gt;Paper and discussion history&lt;/a&gt;. A mention is not an endorsement.&lt;/p&gt;</description><category>cs.CL</category></item></channel></rss>
